[yt-users] installing YT without internet connection

Tue Feb 4 01:10:34 PST 2014

Hi Philipp,

good, I'm glad that it worked!

On Monday 03 February 2014 18:58:18 Philipp Girichidis wrote:
> Looking at your solution I thought it might be easy to include a switch at
> the beginning of the install script and maybe download the yt-folders with
> hg at the beginning (from the local machine). In the lines with the actual
> hg command one would then need a simple copy command to the src/yt-*
> folders (on the remote machine). With this modification it should also work
> without sshfs and the script would run without crashing. I will try to
> modify the script accordingly and keep you updated. 
In general, I tried to keep the changes to the original install_script.sh to a 
minimum, in order to apply them easily, when a new version of the script is 
released. Nevertheless I am very interested to the modifications proposed by 
you; please let me know if you have time to write them down.

Cheers,

  Luigi

> Thanks and best regards,
> Philipp
>  
> Gesendet: Montag, 03. Februar 2014 um 15:50 Uhr
> Von: "Luigi Iapichino" <luigi at uni-heidelberg.de>
> An: "Discussion of the yt analysis package" <yt-users at lists.spacepope.org>
> Betreff: Re: [yt-users] installing YT without internet connection
> Dear Philipp, dear yt users,
> 
> I attach to this email the modified version of my install script, used to
> install yt 2.6 on a machine without internet access. The procedure for the
> installation is also attached as a separate txt file, but I copy it below,
> for sake of simplicity, in case of later discussion.
> 
> As I wrote a few hours ago, this is not straightforward, but it works for
> me, and I haven't come to any better solution. If anyone else has better
> suggestions, I would be glad to hear from them.
> 
> Cheers,
> 
> Luigi
> 
> ---
> 
> Here I describe the installation of yt 2.6. For other versions, please check
> the version numbers of the software that I mention below.
> 
> Let's start with a bit of syntax:
> 
> <remote-machine>: the remote machine without Internet access (a high-
> performance supercomputer, most of the times);
> <remote-username>: your username on that machine;
> <remote-directory>: a directory containing the yt source, in the remote
> machine (create it, if not already existing);
> <local-machine>: a PC with Internet access (your local PC, for example)
> <local-directory>: on your local machine, this is an initially empty
> directory where you will mount the content of <remote-directory>.
> 
> Moreover, every command given below will be preceded by <remote-machine> or
> <local-machine>, indicating on which machine it should be executed.
> 
> First of all, in order to install yt, or to update it by using Mercurial on
> a machine with no Internet access, you have to remotely mount the source
> directory in your local PC.
> So, in order to remotely mount <remote-directory> on <local-machine>, the
> command is:
> 
> (<local-machine>) sshfs -o workaround=rename <remote-username>@<remote-
> machine>:<remote-directory> <local-directory>
> 
> After this command, on <local-directory> you will find the content of
> <remote- directory>. After working with it, you can unmount with
> 
> (<local-machine>) fusermount -u <local-directory>
> 
> Beware: both sshfs and fusermount have to be installed on <local-machine>,
> and in order to use them you should be inserted in some special user group
> (I don't remember which one exactly, but if you ask your local
> administrator it will be fixed for you).
> 
> Now, the yt part: I attach here a modified install script. The most
> outstanding changes with respect to the standard script are:
> 
> - two paths for DEST_DIR, one for <local-machine> and one for <remote-
> machine>. You must switch between the two of them, because the script will
> be executed partly on <local-machine>, and partly on <remote-machine>.
> 
> - INST_HG=0 : you don't need Mercurial on <remote-machine>, but you have it
> already on <local-machine>. Beware: it's somehow funny, but although you
> don't need hg on <remote-machine>, it should be installed there. If not,
> please substite the lines 803-810 with the following ones:
> 
> # We assume that hg can be found in the path.
> # if type -P hg &>/dev/null
> # then
> export HG_EXEC=hg
> # else
> # echo "Cannot find mercurial. Please set INST_HG=1."
> # do_exit
> # fi
> 
> which simply disable a check that would raise an error.
> 
> - Comment the installation of pip (l. 844-845: it does not work on SuperMUC,
> with this installation procedure)
> 
> - Installation of distribute: before starting the script, uncomment the
> lines 841-842. After the install script reaches that point, comment them
> again.
> 
> - l. 882 to 885: I moved there the installation of some required software,
> not included in the initial download step.
> 
> Here is finally the installation procedure:
> 
> 1) (<local-machine>) Perform the remote mount, as described above.
> 
> 2) (<local-machine>) Create a directory called yt-... and copy there the
> modified installation script. Make the script executable.
> 
> 3) (<local-machine>) Execute the script, up to the point where the sources
> are downloaded. The file distribute-0.6.32.tar.gz is not taken together
> with the other ones, It must be added to src manually (important!).
> Moreover, you need two additional source files copied in src:
> pyparsing-1.5.6.tar.gz and python- dateutil-1.5.tar.gz (look for them on
> the Web).
> 
> 4) (<remote-machine>) Execute again the install script, which now compiles
> the tools until Python and then fails.
> 
> 5) (<local-machine>) For going on, the script executes the hg-related
> commands for the yt source. After this part it crashes, and I go again to
> <remote- machine>.
> 
> 6) (<remote-machine>) It will try to update yt from here, failing. Stop
> again and do it from <local-machine>, then go back to (<remote-machine>).
> 
> 7) (<remote-machine>) Go on with the script, which (hopefully!) now proceeds
> to the end without further problems.
> 
> ---
> 
> On Monday 03 February 2014 09:27:31 Luigi Iapichino wrote:
> > Hi Philipp,
> > 
> > I did exactly what you are looking for, on SuperMUC in Garching. I will
> > send my modified version of the install script, with instructions for the
> > installation, in the course of the day. It's a rather cumbersome
> > procedure so, if anyone else has met and solved the same problem, please
> > join the discussion, so that we can compare and improve together.
> > 
> > Cheers,
> > 
> > Luigi
> > 
> > On Monday 03 February 2014 08:47:12 Philipp Girichidis wrote:
> > > Dear YT users,
> > > 
> > > I have a problem during the installation of yt. On the supercomputer,
> > > where
> > > all our simulations are stored, we do not have internet connection
> > > because
> > > of 'security reasons'. So, I tried to download the data on another
> > > machine,
> > > then copy the files to this supercomputer and tried to install
> > > everything
> > > from the packages. But there are quite a few other points in the install
> > > script, where the setup process needs access to the internet. I also
> > > tried
> > > to set up a virtual machine with a linux system similar to the one in
> > > the
> > > computing center, but no success. Did anyone run into a similar problem
> > > and
> > > got a solution or has a guess how to first download everything and then
> > > installing yt without further internet connection?
> > > Best regards,
> > > Philipp
> 
> --
> 
> ---------------------------------------------------------------
> 
> Luigi Iapichino
> Universität Heidelberg, Zentrum für Astronomie
> Institut für Theoretische Astrophysik
> Philosophenweg 12, D-69120 Heidelberg, Germany
> Tel: +49 6221 548983, Fax: +49 6221 544221
> e-mail: luigi at uni-heidelberg.de
> URL: http://www.ita.uni-heidelberg.de/~luigi/
> _______________________________________________
> yt-users mailing list
> yt-users at lists.spacepope.org
> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

-- 

---------------------------------------------------------------

Luigi Iapichino
Universität Heidelberg, Zentrum für Astronomie
Institut für Theoretische Astrophysik
Philosophenweg 12, D-69120 Heidelberg, Germany
Tel: +49 6221 548983, Fax: +49 6221 544221
e-mail: luigi at uni-heidelberg.de
URL: http://www.ita.uni-heidelberg.de/~luigi/